DFI inaccurate temps!

pinktri

Weaksauce
Joined
Jan 18, 2005
Messages
92
Guys, I finally believe that I'm not insane. Here's the story:
My opty 170 was clocked to 2.6 with load temps at 57C idle 37C. I was worried that my temps were too high so I uncapped my proc which resulted in temps at 55C load and 35C idle. Still not satisfied I changed my HS from a Zalman 7700 to a SI-120 and temps were 52C load and 35C idle. Still not satisfied after seeing countless amounts of people with sub 50C load temps I reinstalled my proc about 25 times. yes that's right, about 25, times an entire tube of AS5 down the drain. Still with the same results. Now I believed it was my case and reversed the air flow cut a blow hole and still no change. Now believing it was my temp sensor on my ASUS premium I connected the temp diodes from my coolermaster fan controller that also measures temps and still the same result give or take 2C. So finally thinking it was my board I installed my entire setup on my brother in law's DFI Ultra-D board and BAM my temps were 33C idle and 38-39 load at 2.6 with 1.41 volts.
Hmmmm.....can anyone tell me if it's my ASUS' diode thats reading incorrectly or does DFI seem to read lower temps? On one hand it's hard to believe that my ASUS is off by 14C and since I used my coolermaster temp reader I believe my ASUS board was reading correctly. I never got a chance to use my CM temp reader on the DFI board (what was I thinking)
Anyway, after reading countless sigs and setups I see that everyone who's using a DFI board has very low load temps so I believe they are reading incorrectly. Just my opinion. Does anyone have any input on this matter?
 
i know this sounds stupid but did you ever feel your hsf when it was on? like at load and idle?
 
Vulcanworlds said:
i know this sounds stupid but did you ever feel your hsf when it was on? like at load and idle?

Yeah I've felt it many times when loading and it's definitely warm to the touch. I believe my HS is working fine. All I can say is when I mounted everything on the DFI board my load temps dropped immediately or should I say my load readings dropped by 14C.
 
thunderstruck! said:
Mobo temp sensors are never accurate..
I agree but 14C? That's extremely poor but I guess it's a combo of both mb's reading inaccurately.
 
yea they've never been persice, the olny way to know for sure what your running is to get a thermal probe in there :)
 
Well, ASUS has had a reputation for inaccurate mobo sensors. My A7N8X's chipset and CPU readings were switched even :p

I'm not sure how accurate DFI's sensors are. What I do know is that the CPU's temp fluctuates between 3C usually, like over 5 seconds it will change rapidly by 3C. Right now it's goes from 36-38C with 100% cpu usage. 3C change seems like a lot..
This is using DFI's software, SmartGuardian.

Also, I'm pretty sure the PWMIC sensor is located next to one of the mosfet heatsinks, and that area gets hotter than the actual PWMIC, which is located next to the power connector.
 
Rockaholic said:
yea they've never been persice, the olny way to know for sure what your running is to get a thermal probe in there :)
That probe will only give you a surface temp which will be much lower than the core temp..
 
Why can't AMD and Intel just build a temperature sensor into their CPU's, that would be so much better, but then we would have Intel cheating their temperatures so people will buy their not so hot CPUs...
 
Borgschulze said:
Why can't AMD and Intel just build a temperature sensor into their CPU's, that would be so much better, but then we would have Intel cheating their temperatures so people will buy their not so hot CPUs...
ha...nice cheap shot there. :)
 
The DFI boards read the CPU temp low by 10C on idle and around 14C on load. So, your observations are in line.
 
HeavyH20 said:
The DFI boards read the CPU temp low by 10C on idle and around 14C on load. So, your observations are in line.

wow I never knew this till now. Where were you 1 month ago? Could have saved me some AS5 and a new heatsink. :)
 
HeavyH20 said:
The DFI boards read the CPU temp low by 10C on idle and around 14C on load. So, your observations are in line.
I highly doubt my CPU is hitting 60c on load.
 
HeavyH20 said:
The DFI boards read the CPU temp low by 10C on idle and around 14C on load. So, your observations are in line.
you sure about that? where are you getting this information.
 
ChingChang said:
you sure about that? where are you getting this information.
Well I don't know where he's getting his information from but I can attest to a 14C difference (same proc+heatsink) with an Ultra-D and Premium so that's definitely proof enough for me.
 
Are your fan speeds identical on both boards? If you want to check absolute temperatures you can grab a laser thermometer from radio shack or a comp store and try to find some exposed bit of your processor to target.
 
pinktri said:
Does anyone have any input on this matter?

For what it's worth, with my DFI Lanparty I get low 30's idle and around 57C load on air with two different Opty's clocked/volted about the same. On water my idle temps don't change but my load temps drop to around 50C. With my Venice 3000+ clocked to 2.8GHz at 1.54v it would barely hit 50C load on air but would also hover in the low 30's at idle.

I have no idea if my board tells me lies or not but that's what it reports.
 
I doubt my processor is getting up to 50-55C right now, because when I touch the base of my heatsink it's not even that warm. The only reason it would be that high is if there is horrible contact between the core and the processor's IHS.
 
the IHS contact is something you can't ignore though ;)

Borgschulze said:
Why can't AMD and Intel just build a temperature sensor into their CPU's, that would be so much better, but then we would have Intel cheating their temperatures so people will buy their not so hot CPUs...
they do, it's just the mobo that sucks at reading it

here's something to ponder.
at stock voltage (1.4v), unlidded on water, my 146 runs ~28-30c idle, ~34-36c load
at 1.56v, idle is around 31-33, while load is 36-39c. surprisingly small change in temps eh?

for those of you who know ANYTHING about circuits and how thermal resistors work, i think you can come up with the same theory i have on why the temp change is so much smaller than it should be ;)
 
Back
Top